[Libvirt-cim] [PATCH] [RFC] Transition back to connect_by_classname()

Dan Smith danms at us.ibm.com
Thu Nov 15 18:02:21 UTC 2007


HE> I'm very sorry to tell you, that it does not work with sfcb.

Well, for what it's worth, I can't get it to fail in this way under
sfcb.  However, I can get pegasus to exhibit this behavior *some* of
the time when forceProviderProcesses=true.  It's about 25% of the
time, and only seems to fail on the first connection attempt.  After
this, both connections succeed 100% of the time (AFAICT).

I feel confident that this is a libvirt problem, as I've been able to
reproduce it (very) occasionally with the attached test program.

If I introduce a delay in one of the paths (i.e. KVM) I can't get it
to fail, which indicates to me that it is a race condition somewhere.

I will follow up on this and try to figure out the problem, but I
don't think it should block our conversion to by-classname libvirt
connections.

It would be good if some other people test this patch and report on
its behavior with this wbemcli command:

HE> [heidineu at localhost ~]$ wbemein http://localhost/root/virt:CIM_ComputerSystem

Reports of libvirt version, bitness, etc would be useful.  I'm seeing
this on libvirt-0.3.3 on an x86_64 machine.

-- 
Dan Smith
IBM Linux Technology Center
Open Hypervisor Team
email: danms at us.ibm.com

-------------- next part --------------
A non-text attachment was scrubbed...
Name: parallel.c
Type: text/x-csrc
Size: 753 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/libvirt-cim/attachments/20071115/b3eadada/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 188 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/libvirt-cim/attachments/20071115/b3eadada/attachment.sig>


More information about the Libvirt-cim mailing list